industrial solution for arc welding control system.
1. The ultimate purpose of the introduction of the welding operation is to achieve welds that meet many specified conditions. These conditions are the result of the general system analysis of the technology, which is from the system source-- Welding source- Welding system, weld, operating system and control system. This paper presents an industrial solution for welding process. Arc voltage of neural network. This result can be applied to automated robotic welding based on the correction of the trajectory welding torch through the weld. 2. Using the previous research, we propose a wig welding control system with Neural Network Controller (Haykin, 1999). For the welding head position control, we use the acquisition system of the arcvoltage value during the welding process. Therefore, the arc voltage is a way to measure the position of the welding head. This signal will be the input of the neural network structure specially made by the author to modify the position of the welding head. The device can realize the real-time control of the weld sensor algorithm. - Two interconnected adaptive linear filters: one filter works on the principle of interference compensation, which extracts useful signals from welding process signals, and the second filter eliminates the reference signal and the first filter outputs the signalBucur etal. ,2008). The industrial welding head control system shown in Figure 1 contains: * welding source. * Filters for processing collected signals The welding arc voltage passes through the filter. * Process- Computer Interface AX 5411. * Signal amplifier. * DC engine (MCC) Used for oscillating motion of welding torch and correction of welding head position. * Reference signal generator. * Synchronous block (BS) In order to synchronize the reference signal with the processing signal, the moment when the processing data starts is obtained. * Personal computer PC, which implements the collection of voltagevalues through AX 5411 interface, processes them through the same interface and generates command signals for the DC engine. [ Figure 1 slightly] The welding source is also controlled by PC through serial portRS 232. The software consists of A/N conversion for obtaining information, A numerical command generation and an N/A conversion program, and A graphical interface with AX 5411 and welding sources (Bucur etal. ,2002). In the intermediate plan of the weld, only the position correction through the position correction of the weld and the welding torch are carried out, and the speed control is carried out by the welding torch. 3. Problems in the welding process during the processing of data using a neural network, there may be some inconsistency between the reference signal and the processing signal, which determines the low quality of the welded joint ( Phenomenon of \"beating)(Miclosi et al. ,1984). Therefore, it is necessary to start processing the data at the moment when the two signals coincide. With this mentioned, we propose the sine signal structure as shown in figure 2. This sinchronization block (BS) Is a comparison element (CSC) An analytical calculation is performed between the arc voltage value and the reference signal. After comparison, at the time of coincidence, this generates a signal to start processing the data ( Dumetescu and Buku, 2000). After this sync, we can demonstrate that the neural network controller realizes the welding torch position correction in the horizontal plan, regardless of these frequencies, and regardless of these coincident moments. [ Figure 2: Two signals of different frequencies are shown in figure 3. The neural network simulation is realized by MATLAB program ( Kharab & Guenther, 2002). The program is: time = 0: 0. 05:10; x= (figure 3 --green courve)t=x+p\'; [w,b]= [a,e]= [w,b]= [a,e]= plot(time,a,time,x); plot(time,e); (figure 4) In this program, x is the reference signal, p is the process signal, w and B is the adjustment weight, a is the output signal of the neural controller, and e is the error signal. The simulation time is 10 seconds. We can observe that the signal is synchronized, and our neural network can realize frequency correction. The error signal obtained during the synchronous process is shown in figure 4. This error is only the amplitude error, not the frequency error. Therefore, we can say that our neural controller produces an output signal, which is a correction signal for the position of the welding head and a correction signal for the movement of this tool. These results are available when the frequency of the signals is different and the coincidence between them rises. We perform other simulations of signals of different frequencies and overlap on the opposite steepness. [ Figure 3 slightly][ Figure 4 slightly] The results are the same: position and frequency correction with similar error values. 4. A technical scheme for wig welding control system is summarized. This industrial system, in the middle plane of the weld, contains only the position correction and welding torch through the weld, without the advance speed control of the welding torch. During wig welding, if the oscillation frequency is modified to the reference frequency of the neural network, we simply prove that the designed network can also correct the oscillation frequency of the welding gun. All these considerations are [delta] = 0, where 5 is the welding of the torch Chiwei from the plane in the seam. It will be very interesting. delta][not equal to] 0, then determine the function [delta] The monotony of the function is studied. We can verify whether our neural network can implement the proposed correction. In the future welding process, the study of neural network control method for welding process is very meaningful. During MAG welding, there will be drop-burning transfer through the welding arc. 5. Reference Bucur, G. ; Popescu, C. & Popescu, C. (2008). Neural network control of wig welding process, Katalinic, B . 19 International daaam seminar record(Ed. ), pp. 84-85, ISBN 978-3-901509-68- 1. Austria, October 2008, published by DAAAM International, Vienna Buku, G. San Dumi treescu& Miclosi, V. (2002). Neural network control of robot welding process. Seminar magazine \"35 deani de active it University gasoline- Gaze at la Ploiesti \", Volume 1. LIV, No. 2/2002, pp. 38- 43, ISSN 1221/9371 dumetescu, St. & Bucur (Chiriac), G. (2000). Petroleum measurement technology University of natural gas Press, ISBN 973-99506-5-5,973-99506-7- 1. Proytis, Haijin, Romania(1999). Neural network, Prentice Hall, ISBN 70-9013-273350-1, U. S. A. Kharab, A. & Guenther, R. (2002). Introduction to Numerical Methods in MATLAB, Chapman & Hall/CRC, ISBN 1-58488-281-6, U. S. A. Miclosi, V. , Scorobetiu, L. , Jora, M. & Milos, L. (1984).